Banks are trying to navigate a tricky balance when it comes to AI adoption. Move too slow, and risk being overtaken by more nimble rivals—but move too fast, and one mishap could destroy one’s reputation as a responsible financial actor.
Craig Corte, global head for digital, data and coverage platforms for corporate and investment banking at Standard Chartered, said he is fine if his employer decided to be a “good follower” on AI, given the risks involved if a major financial institution screws up.
“I don’t think we should be at the cutting edge of innovation around AI as a big bank. I think that’s a risky place to be, and there are a lot of other organizations and industries that can be there,” Corte said last week at the Fortune Brainstorm AI Singapore conference. “I think we can afford to be a little bit behind the curve.”
Tianyi Zhang, general manager of risk management and cybersecurity at Ant International, pointed to three risks posed by AI. (Ant International is a partner of Fortune Brainstorm AI Singapore.) The first is AI’s penchant to make things up, or “hallucinate.” The second is the possibility for different AI agents to work directly with each other, which opens up new avenues for external attacks. The third is deepfakes, including the possibility that fake customers are generated as an attack vector.
Still, Zhang said AI was making parts of his job easier, offering up the example of how it can augment the skills of entry-level financial investigators.
Banking customers are also thinking about whether to trust AI. Vivien Jong, chief digital and AI officer for Asia at BNP Paribas Wealth Management, noted that younger clients have embraced AI due to its speed and transparency. “They want to use AI to look for thematic investing around sustainability or tech,” she said. Older customers, however, are more cautious, seeing the new technology as a “support tool, and not something to be used for investing.”
Large vs. small
Corte and Zhang were part of a panel exploring how AI is set to transform the financial industry. One key question was what kind of institution might benefit most from AI: large established banks, or smaller scrappier startups?
Larger established players have previously been slow to adopt new technologies—and often paid the price for their hesitation. But this time around, bigger companies are far more eager to adopt AI.
“For those of us that were around in the first digital revolution, it was a bunch of outsiders and small companies trying to convince the big incumbent players that they needed to digitize their businesses,” Corte said.
But unlike previous instances of digital transformation, where larger established players struggled to keep up, bigger banks are more eager to adopt new technology this time around.
“For those of us that were around in the first digital revolution, it was a bunch of outsiders and small companies trying to convince the big incumbent players that they needed to digitize their businesses,” Corte said. “That is completely reversed today. The biggest players in the world with the most customers, with the biggest balance sheets, [they] are the ones driving the AI agenda.”
Smaller startups, meanwhile, can struggle with long-term horizons or lengthy documentation needed to work with a big bank. Jong, from BNP Paribas, shared her own struggles about working with smaller startups, including one that “went offline because it didn’t get paid for two weeks.” One hangup was the size of BNP’s contracts. Jong recounted that one startup was so uneasy about a 60-page master service agreement, it said it would rather work for free for six months.
Zhang, from Ant, approached the conversation of size from a different vantage point: Ant’s customers.
“Some of our clients…are very small. They could be a couple, a husband and wife operating their online store in their one-bedroom apartment,” Zhang noted. Normally, such small customers would struggle to handle all the different risks that come with running a small business. But “with AI’s help, they can have access to all the new technology, new tools to deal with automated payments. They can deal with dispute solutions, risk management solutions, and they can collect money from different currencies and deal with foreign exchange volatility,” he said.
AgentFi
Michael Wu, CEO of crypto firm Amber Group and a speaker on last week’s panel, is all-in on how AI can shake up the financial sector. Amber is now pursuing “AgentFi,” or finance driven by AI agents that can autonomously make their own decisions. (Disclosure: Fortune’s owner, Chatchaval Jiaravanon, is an investor in Amber Group)
Wu noted that AI agents currently don’t have the financial resources to carry out the actions they decide to take. “An agent cannot have the autonomy to say ‘hey, I want to spend this amount of money, or I want to invest in this versus that,” he said.
Crypto, Wu argued, will give AI agents “financial freedom,” and give them the resources to put behind their decisions. “They could even hire humans back to do what they want,” he suggested.
Amber launched its first “agent,” an AI dubbed “Mia,” to serve as the group’s “AgentFi Ambassador” in May. “My best analogy is [that Mia is] a very bright, young, super intern,” Wu said. “She can do some things amazingly. She still makes a lot of mistakes, and sometimes she behaves very dumb, to be upfront.”
Wu’s engineers gave Mia the ability to manage the liquidity of its own token. Yet, Wu noted the agent struggled to describe what financial actions it was taking on social media. “It happens to humans too, right? Sometimes we learn a new thing very quickly, and our left or right brain…doesn’t realize what the other half is doing.”
“Hopefully, this time next year, a lot of these engineering problems will be spotted, identified and potentially solved by these agents themselves,” he added.
